The computer-made system, known as NASNet, is designed to identify objects, such as people and vehicles, in photographs and videos, the search engine giant says.
Studies show that NASNet is able to identify objects in an image with 82.7% accuracy. Google says this is an improvement of 1.2% over AI programmes created by humans.
The web giant has made the system “open source”, which means developers from outside the company can either expand upon the programme or develop their own version.
Researchers at Google say they hope AI developers will be able to build on these models to address “multitudes of computer vision problems we have not yet imagined.”